Answer frequent questions
From your information and documents; if it is not sure, it hands over instead of improvising.
WhatsApp AI chatbot connected to your business
Digitup Studio is a technical studio in Madrid that develops WhatsApp AI chatbots on the official WhatsApp Business API: they answer with your company's information, query your CRM, calendar or ERP and hand the conversation to a person when needed.
In short
- A WhatsApp AI chatbot answers your customers in natural language using a language model and your company's data.
- It is built on the WhatsApp Business Platform, Meta's official API: the WhatsApp Business app is not enough to connect it to your systems.
- It reads and writes in your CRM, PMS, ERP or calendar, and hands the conversation to a person with a summary.
- The cost has three layers: development (from €1,000), Meta messages and AI model usage.
- When it fits, we start from AlisaBot, our own chatbot platform for WhatsApp, Telegram, web and WordPress.
Features
What can a WhatsApp AI chatbot do?
Resolve the repetitive conversations and prepare the rest for your team, using real data from your business.
Availability and bookings
Checks the calendar, PMS or catalogue in real time and creates or changes the booking.
Recommendations and sales
Suggests products or services, qualifies the contact and passes it to the sales team.
Issue handling
Collects the issue, classifies it by type and urgency and logs it for the right team.
Classification and summaries
Tags each chat by topic or urgency and summarises the thread for whoever picks it up.
Notifications
Confirmations, reminders or status changes, using templates approved by Meta.
For a website or Telegram, see AI chatbots and assistants. If it has to run tasks across several systems, you need AI agents; for a simple WhatsApp notification, an automation is enough.
Approach
Why a custom chatbot and not a generic tool?
Because the value lies in what the chatbot can look up and do in your systems and with your rules, and that does not come in any template.
If all you need is an FAQ bot, an off-the-shelf tool is a good option. It falls short when the answer depends on a PMS, an in-house CRM or a specific business rule. We do not just plug in ChatGPT: we design flows, memory, context, data, prompts, rules, integrations and control panels. When it fits we start from AlisaBot, our own platform, and build what is specific to your business on top of it.
App or API
What is the difference between the WhatsApp Business app and the API?
The app is free and built for small businesses that reply from a phone; the API (WhatsApp Business Platform) connects an AI chatbot to your systems and Meta charges for it per message.
| Aspect | WhatsApp Business app | WhatsApp Business Platform (API) |
|---|---|---|
| Who it is for | Small businesses (according to Meta) that reply by hand from the app. | Companies that handle conversations at scale from their own software or a provider's. |
| Price | Free to download. | Meta charges for each message the business sends, depending on the category and the recipient's country. |
| Automation | Greeting and away messages and quick replies. | Whatever you program: AI, rules, integrations and handoff to people. |
| Connection to CRM, calendar or ERP | Not directly: it requires moving the number to the API, for example with coexistence. | Yes, through a backend connected to your tools' APIs. |
| Volume | As much as your team can handle. | Up to 80 messages per second per number, by default. |
Meta sources consulted on 23 October 2026: WhatsApp Business app, platform overview, pricing and coexistence. Meta changes these terms often.
Requirements
What do you need to get started with the WhatsApp API?
A Meta business portfolio, a phone number of your own, approved templates for the messages your company initiates and your customers' consent.
- Meta business portfolio. It is mandatory. Verifying the business is one of the ways past the initial limit of 250 unique users every 24 hours on messages the business initiates.
- A number of your own. With a country prefix (short numbers do not work) and able to receive the code by SMS or call. If it is already on WhatsApp, that account is deleted or coexistence is used.
- Approved templates. These are the messages your company initiates, such as reminders or confirmations. They are classified as marketing, utility or authentication, and Meta reviews them before they can be used.
- Consent. Before messaging someone on your company's initiative, Meta requires their opt-in, with the company name clearly stated.
You provide the content, access to your systems and the handoff criteria; we help with the technical setup and build the backend, integrations and panel. Number requirements: Meta documentation.
Integrations
How does it connect to your CRM, calendar or ERP?
Through a backend that talks to WhatsApp and to your tools' APIs, and applies your rules before reading or writing anything.
The model interprets the message and asks the backend for the action, which checks permissions and runs it; it never touches your systems directly. We usually build it with Python/FastAPI or Node.js/NestJS on PostgreSQL or MongoDB. What gets connected most often:
- CRM, including a custom CRM.
- PMS and channel managers for accommodation.
- Appointment calendars and scheduling tools.
- ERP, Google Sheets, databases and in-house or third-party APIs.
If your ERP works, it stays: we build the missing part. To answer from documents we use RAG for business, which updates without retraining models. To bring AI into software you already have: AI integration. We apply access control, data separation, action logs, credential protection and backups.
Handoff
When does the conversation go to a person?
When the customer asks, when the chatbot is not sure or when a rule flags it, such as a complaint, a payment or sensitive data.
We define with your team the reasons for handoff and who receives each case. That person gets the thread summarised and classified by topic and urgency, and steps in from the agent panel; the review panel is used to audit and improve the answers.
Each customer message opens a 24-hour service window; outside it only approved templates can be sent, so the team's working hours and the handoff rules are designed together. AI finds, classifies, summarises and prepares; the person reviews, decides and approves.
Price
How much does a WhatsApp AI chatbot cost?
Developing it with Digitup Studio costs from €1,000; on top of that there are two usage costs we do not set: Meta messages and AI model consumption.
| Layer | Who charges it | What it depends on |
|---|---|---|
| Development | Digitup Studio, from €1,000. | What the chatbot does, on which channels, which systems it connects to and how much information it handles. |
| WhatsApp messages | Meta, per message the business sends. | The message category (marketing, utility, authentication or service) and the recipient's country. |
| AI model | The model provider, by usage. | The chosen model, the volume of conversations and the context sent. |
A simple chatbot with your company's information is closer to the minimum; integrations with a CRM, calendar or ERP raise the price.
Meta has charged per delivered message since 1 July 2025, and only for messages the business sends. From 1 October 2026 it also charges for service messages, with 1,000 free per month per number. Per-country rates are on its official pricing page; we do not copy them because they change.
Model consumption is controlled with usage limits, monitoring and a model suited to each task. Depending on the project, development is paid as a fixed price, over 6 or 12 months, or with Build & Run, a monthly fee with AI and WhatsApp limits agreed in advance; we detail it in ways of working.
Regulation
Which regulations affect a WhatsApp AI chatbot?
The EU AI Act requires telling people they are talking to an AI, the GDPR applies because personal data is involved and Meta has its own policies.
AI Act. Article 50 of the EU AI Act, applicable from 2 August 2026 according to its Article 113, requires that people know they are interacting with an AI system, unless it is obvious. The usual approach is to say so at the start of the conversation and offer to talk to a person.
GDPR. You need a legal basis, you must inform people, keep only what is necessary and check where model and messaging providers process the data.
Meta policies. They require each person's consent before you message them on the company's initiative and treat providers of general-purpose AI assistants separately, with a policy of its own.
We do not give legal advice: validate this part with your adviser or your data protection officer.
Which chatbots has Digitup built?
The WhatsApp integration lives in AlisaBot, our own product. Inferno IA, built on AlisaBot, and the Soñar Despierto chatbot plugin are web chatbots.
Fit
When does a custom WhatsApp chatbot fit and when not?
It fits when WhatsApp already has volume and the answers depend on your data; not if an automatic message from the app is enough.
When it fits
- You receive many repeated enquiries on WhatsApp.
- Answers depend on data that changes: availability, orders, appointments or statuses.
- Several people handle the same number and you need handoff and history.
- You already use a CRM, PMS, ERP or calendar and want the chatbot to work with them.
When not
- The greeting and away messages of the WhatsApp Business app are enough for you.
- An off-the-shelf tool, already configured, covers what you need.
- The budget is below €1,000 or delivery has to happen in under 2 weeks.
- You want to message people without their consent: Meta does not allow it.
Not sure where to start? The AI audit prioritises the case with the best return, and we will tell you if we think it is not worth doing.
Related services
What usually goes alongside a WhatsApp chatbot.
Frequently asked questions about WhatsApp AI chatbots
Can you put an AI chatbot on WhatsApp?
Yes. It is done with the WhatsApp Business Platform, Meta's official API; the WhatsApp Business app does not connect to your systems on its own. Messages reach a backend, a language model interprets them using your company's information, and the chatbot replies, queries your CRM or calendar, or hands the conversation to a person.
How much does a WhatsApp AI chatbot cost?
Development with Digitup Studio starts at €1,000: a simple chatbot with your company's information is closer to that minimum, and integrations with a CRM, calendar or ERP raise the price. Separately you pay for Meta's messages, depending on category and country, with 1,000 free service messages per month per number, and for AI model usage.
Can I use my current WhatsApp number?
Usually yes, if the number is yours, has a country prefix and can receive the verification code by SMS or call. If you already use it on WhatsApp, you can delete that account before registering it on the API, or use coexistence, which keeps the app for one-to-one chats, although it disables some features and leaves broadcast lists read-only.
Can the chatbot hand the conversation to a person?
Yes. It hands over when the customer asks, when the chatbot is not sure or when a rule flags it, such as a complaint, a payment or sensitive data. The person receives the thread summarised and classified by topic and urgency, and replies from an agent panel. After 24 hours since the customer's last message, they can only write with a template approved by Meta.
Does it connect to my CRM, calendar or ERP?
Yes. The AI model does not touch your systems directly: it asks a backend for each action, which checks permissions and calls the APIs of your CRM, calendar, PMS, ERP, Google Sheets or database. That way it can check availability, create a booking or log an issue applying your rules. If your ERP works, it stays: we build the missing part.
Can it message customers who have not written first?
Yes, under two Meta conditions: the person must have agreed to receive messages from your company, identified by name, and the message must be an approved marketing, utility or authentication template. In addition, a new business portfolio can only message 250 unique users every 24 hours on its own initiative; the limit rises, for example, once the business is verified.
Do you have to disclose that it is a chatbot?
Yes. Article 50 of the EU AI Act, applicable from 2 August 2026, requires that people know they are interacting with an AI system, unless it is obvious. On WhatsApp the usual approach is to say so in the first message and offer to talk to a person. The GDPR also applies: inform people about the processing and keep only the data you need.
How long does it take to get it running?
Digitup Studio projects are delivered from 2 weeks. A first working version can usually be validated within weeks if the scope is clear; complex integrations with a CRM, PMS or ERP may need more phases. It is worth preparing the Meta side in parallel: business portfolio, number, display name and templates, because templates need Meta's approval before they can be sent.
Which AI model does the chatbot use?
The one that best fits each task. We integrate models from OpenAI, Gemini, Anthropic, DeepSeek or Perplexity depending on quality, cost, speed, privacy and type of task. A single chatbot can use a fast, inexpensive model to classify messages and a more capable one to draft replies. Consumption is controlled with usage limits and monitoring, and it is a cost separate from development.
Where is Digitup Studio based?
Digitup Studio is a technical studio founded in 2019, based at Calle Alcalá 211, Madrid. We work mainly remotely with companies in Madrid, the rest of Spain and other countries, and meet in person when the project calls for it. We serve clients in Spanish, English and Russian. You can start with a 30-minute call with Anton Zverev, our CTO, with no commitment.
Shall we talk about your WhatsApp chatbot?
Tell us which questions you receive on WhatsApp and which systems you work with. In a 30-minute call with Anton, our CTO, we will tell you how we would approach it, or whether it is not worth it.